1

DESIGNING A SOCIAL SECURITY PENSION SYSTEM

Robert L. Brown Professor

Department of Statistics and Actuarial Science University of Waterloo

Waterloo, Ontario, Canada N2L 3G1 Phone: (519) 888-4567

rlbrown@uwaterloo.ca

ABSTRACT

This paper looks at potential models of Social Security systems. It refers often to the systems that exist in the United States and Canada (the latter more particularly) to outline the issues involved in attempting to design a “good” social security pension system. Of course, one of the issues is the definition of “good”. This paper will use criteria such as: poverty alleviation, retirement income adequacy, benefit/contribution sustainability, income equality and wealth distribution. In the course of the discussion, the reader will be exposed to many issues that need to be addressed in the establishment of any Social Security system in the world. This may prove to be helpful in countries where new systems are established (this could be any country in the world, since even existing systems are always evolving). It is also hoped that future students of Social Security will find this paper helpful in that it is meant to lay out some basic principles consistent with good Social Security design. Keywords: Social Security Pay-as-you-go Individual Accounts Notional Defined Contribution

2

I Introduction What makes a good Social Security system? It is the goal of this paper to outline in general terms a number of key principles that need to exist in any Social Security system that would earn a “good” review. The paper describes Social Security in a fairly narrow sense. It does not include the provision of Health Care delivery nor Workers Compensation. Further, the paper uses the Social Security systems that presently exist in Canada and the U.S. as benchmarks against which one can consider issues of optimality. Thus, the paper will explore issues related most particularly to the provision of Retirement Income Security by government-sponsored systems. II Competing Views of Social Security Pension Design The ultimate design of any social security system is the responsibility of public policy makers (usually politicians). These policy makers often use economists and/or actuaries for advice. More recently, the economists at the World Bank have had some significant influence on the reform of social security systems in many countries. The involvement of actuaries is normally reserved solely to “pricing” the various designs brought to them by the policy makers. Why is this an issue? Why would one care whether Social Security systems are influenced by economists versus actuaries? According to the International Actuarial Association (IAA), economists and actuaries come to the drawing table where social security systems are designed with a different set of principles and priorities. The following arguments can be found in a recent document (IAA 2006) in which the IAA Social Security Committee responded to the World Bank (2005) publication: Old-Age Income Support in the 21st Century: An International Perspective on Pension Systems and Reform. In their response, the IAA states (p 3/4):

“We consider that actuaries and economists view Social Security systems in remarkably different ways.

When considering these models, it appears that economists have as their

priorities: • individual equity • reduction in labor force distortions caused by social security (contributions and

age of retirement) • national savings • strong financial institutions • wealth creation

3

Some economists believe this naturally leads one toward a Defined Contribution model.

… By comparison, actuaries tend to focus more on: • insurance and risk sharing (based on the law of large numbers) • annuitization for the longevity risk (or other similar mechanisms) • low expenses • a long-term view of stability and sustainability (not an annual balancing

requirement) • predictable benefits • improvement in societal utility (of wealth) • transparency and understandability This focus on insurance, risk-sharing, and the law of large numbers, leads

more naturally to Defined Benefit national social security models. Although some of the above can be obtained through DC systems, it is more difficult to maintain the above principles for all time, as can be seen from United States benefits history. For example, individuals with individual accounts who see their account balance every quarter will be more likely to demand access to them, and resist required annuitization (as experience has shown in the United States with IRAs and 401(k) arrangements).”

In this regard, one might ask whether the new World Bank mantra of Notional Defined Contribution plans for Social Security are Defined Benefit or Defined Contribution Plans. Michael Cichon (1999 and 2005) argues (convincingly) that the “new” Notional Defined Contribution plans are mathematically equivalent to Career Average Defined Benefit Plans if the Notional Investment return on the former is equal to the Salary Scale function in the latter. U.S. actuaries will recognize Notional Defined Contribution Social Security as a version of Cash Balance Plans which are categorized as Defined Benefit Plans in the U.S. because the benefit is “guaranteed” (more on this in a moment) and the major plan risks (e.g., investment risk, interest-rate risk and longevity risk) can be borne by the pension plan sponsors (in the case of social security, the plan sponsor is society). Economists designed these Notional Defined Contribution schemes because they satisfy most of the priorities of economists listed above. Actuaries might find some comfort in these Notional DC schemes since they are equivalent to Defined Benefit Career Average plans. Actuaries have objected strenuously, however, to the automatic balancing formula proposed in the Swedish Notional DC system (see Scherman, 2005 and Hagberg and Wohlner, 2002) which means the benefits are not guaranteed.

4

As for Individual Account Social Security schemes, there are other (actuarial) reasons to oppose them. As we know, these systems transfer virtually all the major risks (investment risk, interest-rate risk (at the time of annuity purchase), inflation risk and longevity risk) to the worker. This is contrary to the actuarial principle of grouping risks and achieving the advantages of the Law of Large Numbers. Small account balances face higher expense ratios. Thus, these systems are regressive. Total expenses in these systems are higher than most government-administered systems. Overall, DC plans do provide retirement savings and may even encourage saving for retirement. However, a savings plan does not a retirement system make! Actuaries tend to favor Defined Benefit designs for all the risk mitigating reasons listed above and the paper will not consider Individual Account Social Security systems further in this discussion of designing an optimal social security system. III Financing Extremes: PAYGO versus Fully-Funded It is the position of this paper that the financing of any social security scheme is of secondary importance. This is based on another fundamental belief that Social Security should not be thought of as a big private pension plan, but rather as a wealth transfer scheme. Under PAYGO financing, the required contribution rate, C = Pt * Bt At AIWt where: Pt = The number of pensioners At = The number of active workers (in the formal sector) Bt = The average pension benefit AIWt = The Average Wage upon which contributions are made. This can be viewed as being the product of the Demographic Ratio and the Financial Ratio. So in a country that tries to replace 39% of income through Social Security benefits for the average worker and where the Aged Dependency Ratio is 0.33, the contribution rate will be 13% (this approximates the U.S. reality quite closely). If we change the formulation slightly, then we can say that for a $1 retirement benefit starting at age 65, the required PAYGO contribution rate at a given point in time is:

5

!

C =e"rx

65

#

$ Lxdx

e"rx

20

65

$ Lxdx

Where: Lx = Population Alive Aged x rx = The growth rate in covered earnings which in turn is dependent on the growth rate of the labor force plus the growth rate in real wages due to gains in productivity. In a Fully-Funded, Individual Accounts Social Security system, again using age 65 as the retirement age, the contribution required per unit of benefit is:

!

C =e"#x

65

$

% lxdx

e"#x

20

65

% lxdx

Where: δ = the rate of return on investments (here assumed constant) lx = Life Table “survivors” which provides survivorship probabilities One sees that these formulae are not remarkably different. Hence, the method of financing Social Security results in remarkably similar economic impacts. In a PAYGO system, workers deny themselves current consumption and make contributions to the social security plan. These contributions are immediately paid out to retirees who use these dollars to buy current consumption. An analogy helps. Assume the social security contribution rate is 10% (who pays it may not matter since the impact will ultimately fall back onto the workers’ shoulders regardless). In this system, workers give up one-half of one days’ worth of product and provide that output to the elderly for consumption (e.g., Monday morning in a five-day week). In a fully-funded system, workers deny themselves current consumption and buy assets from the marketplace (they invest). Years later when these workers retire, they then must find current workers to buy their assets so as to transfer the value into currency to purchase consumption. However, the end result is virtually identical. Both systems are absolutely dependent on a next generation of workers to produce goods and services. Neither is demographically immune. It does not matter a whit how much money you have; if no-one is producing Gross National Product, you will starve to death in the dark.

6

Many proponents of fuller funding (or individual accounts) argue that such systems are more stable than PAYGO systems. However, neither system is particularly stable. At the very least, neither system is guaranteed to be more stable than the other. Which is more stable: interest rates or fertility rates? PAYGO financing depends strongly on the Demographic Ratio. No one can really suggest that it can be predicted 25 to 50 years out. The fully-funded system depends heavily on the rate of return on invested assets. No one can really suggest that these rates can be predicted 25 to 50 years out (or even their average value over that time period). Both systems suffer from political risk. In a PAYGO system, the government can change the benefit or contribution formula and deny workers their assumed contract. In a fully-funded system, the government can allow inflation to deplete the value of the Individual Accounts, once again nullifying the implied contract with workers. For a fuller discussion see Brown, 1997, Barr, 2000 and Conesa and Garriga, 2004. Further, in a fully-funded scheme, the local dictator can abscond with the plan assets in the middle of the night. With a PAYGO plan, he can only steal the plan liabilities! Beyond this point, the aspect of financing will not be of primary focus. The critical ultimate outcome in the affordability of any social security system is a healthy and growing economy. The method of social security financing may be close to irrelevant in this regard. IV Issues in the Design of a Social Security Pension System What should be the priorities in any well-designed Social Security pension system? It can be argued that the number one priority should be the mitigation and alleviation of poverty amongst the elderly. If one accepts this as the primary priority then it follows that an optimal social security system will transfer more wealth to poor participants than to wealthy participants. A second goal is to help citizens maintain an acceptable standard of living post-retirement. This means that there will exist some notion of an acceptable “replacement ratio” provided by the Social Security system. But note that the goal is to help citizens in attaining an acceptable standard of living. The Social Security system does not need to be the sole provider of such security. Most countries design their Retirement Income Security systems as a combination of: government-sponsored systems, employer-sponsored plans and individual savings. These broader-based systems often come with significant tax incentives (and thus are at least partially “public” systems). For Canada and the U.S., within defined limits, contributions to qualified plans (employer or individual) are tax deductible (i.e., come out of before tax dollars), the investment income accrues tax free and income, when taken, is taxable in full.

7

A third goal is “solidarity”. That means that all contributors (normally workers and employers) should want to support the Social Security system. That is, there should not be a large proportion of workers who do not participate and benefit in the system. This will mean that there will have to be benefits for wealthy people even if that may seem otherwise unnecessary. One also needs to be wary of providing significant minimum benefit guarantees in the total package of benefits. This may seem to be consistent with the priority of alleviation of poverty, but it brings with it huge risk of moral hazard. That is, if workers (and employers) know that by achieving a minimal hurdle one can achieve enough benefits to avoid poverty, many workers (abetted by their employers) will just achieve that hurdle and not one millimeter more. This will mean that the system will pay benefits for which there are not equivalent contributions. At the least, such benefits should be minimalist (only alleviate poverty) and come from a program which is financed from general tax revenues, not Social Security contributions. More on this later. To the extent possible, the Social Security system should not create perverse economic incentives. These include incentives to stay out of the formal economy (and pay taxes and Social Security contributions) and enter the cash economy. That means that the total of taxes and Social Security contributions cannot be too high. What “too high” means will vary from time to time and culture to culture, but there is definitely a limit on the total of taxes and contributions beyond which a country will reap fewer dollars of income. Similarly, one does not wish to have the total of taxes and contributions be so high that employers do not want to hire new labor. Employers have a choice. If taxes and contributions get too high, they can move work offshore by outsourcing. Even before acting to that extent, employers have the choice of offering overtime to existing workers rather than hiring new workers. Depending on the design of the Social Security system, offering overtime to existing workers may not add any extra Social Security contributions as most systems have a maximum salary to which contributions attach. The system should not have a benefit structure that creates any disincentives for individual workers also to save for their own retirement. This could be the result if your benefits are too large. But for marginal workers this could also result if there exists a “claw back” of early public benefits that is so rapid (steep) that it creates the equivalent of very high marginal tax rates on personal savings. The system design should not create an incentive for workers to leave the workforce prematurely. This can happen if workers get no increase in benefits once they have contributed for “n” years (which is true in both Canada and the U.S.). It is also true if the adjustment for early or late benefits is not a full actuarial adjustment (again true in both Canada and the U.S.).

8

Finally, the system design should not encourage unnecessary absences from the labor force. This can happen if you have benefit formulae that exempts periods of time out of the labor force for certain defined reasons. The Child-Rearing Drop Out in Canada is such an example. Exemptions from periods out of the labor force because of disability exist in both Canada and the U.S. However, it is the position of this paper that it is better by far to have explicit drop-out provisions (as in Canada) than to just have a short qualifying period for benefits (say, 20 years) for all. Any social security system will also need to have sustainable contributions and benefits. It is somewhat obvious that it is impossible to create any Social Security system that satisfies all of these priorities. Some economists would argue that Defined Contribution plans come the closest. Many actuaries would argue in favor of Defined Benefit plans. However, all of the above criteria need to be considered in the design of a country’s system. V Two Living Examples: Canada and the U.S. Given the priorities outlined in Section IV, the paper now looks at the systems in Canada and the US to see how they attempt to satisfy these goals.

The optimal design features presented in this paper are produced in a context of a country with an honest government and good governance. That is, the proposals presented are theoretically optimal given the validity of these assumptions. All listed criteria would have to be reconsidered in a country with a dishonest government or with weak governance. As will be seen, both Canada and the U.S. come close to satisfying the listed criteria. However, at this moment, both OASDI and the QPP have some problems with their long-term sustainability (neither plan is in actuarial balance projected out 75 years). In the 1996 amendments to the Canada Pension Plan (CPP), a sustainability clause was added to the scheme. Given the 1996 amendments, the CPP is meant to exist forever with a contribution rate of 9.9% of Pensionable Earnings (the contribution is split 50/50 between employers and workers, while the self-employed pay the full 9.9%). If, in any actuarial valuation, the plan actuary finds that the CPP cannot be sustained for 75 years with the 9.9% contribution rate then a two-pronged response automatically occurs. First the contribution rate is raised so as to cover one-half of the long-term deficit. Second, benefits are frozen (do not rise with indexation) until the other half of the deficit disappears. Thus, the immediate response is shared virtually 50/50 between contributors/workers and beneficiaries/retirees. This seems preferable to the new Swedish automatic balancing mechanism where all of the response comes from decreasing benefits (through de-indexation) until equilibrium is satisfied (i.e., contributions must remain constant). This feature of the new Swedish system has been

9

highly criticized (Schermann, 2005 and Hagberg and Wohlner, 2002) and means that Swedish benefits cannot be considered “guaranteed”. It is the paper’s position that an automatic stabilization feature is highly beneficial and that the Canadian methodology could be used as a suitable model. The paper will now discuss the apparent “goodness” of the social security systems used in Canada and the U.S. based on other criteria previously listed.

The following six figures give a pictoral image of the manner in which these two countries provide retirement income security through government sponsored systems. Because the total benefits in both countries depend on the amount of personal income of the individual worker, the following graphs assume that the worker will save enough so that his/her private retirement income plus total social security benefits will provide a replacement ratio of 70% of final earnings. If the government social security benefits provide this 70% replacement ratio, then the worker is assumed to have saved nothing and provides no extra retirement income. The Canadian Social Security system has three tiers. The Guaranteed Income Supplement (GIS) is a welfare benefit. It is paid depending on a person’s income from the previous tax year (data come from Income Tax filings). There is no asset test. The benefits are paid from general tax revenues. GIS Benefits are “clawed back” at the rate of 50% for each dollar of personal income (except OAS). GIS benefits are non taxable. Old Age Security (OAS) is usually referred to as a “Demogrant” benefit. You must have 40 years of residency between ages 18 and 65 to get a full OAS (fewer years result in pro rata benefits). However, the OAS benefit also has a claw back whereby 15% of one’s OAS is lost for each dollar of personal income once your income exceeds a stated threshold (see http://www.sdc.gc.ca/en/isp/pub/factsheets/rates.shtml. for details). Thus, wealthy Canadians receive no GIS nor OAS. The OAS claw back is based on individual income whereas the GIS claw back is based on household income; thus some elderly females (normally it is females) in wealthy households still get the OAS. OAS benefits are taxable income. The Canada/Quebec Pension Plans (C/QPP) are virtually identical and provide retirement benefits (70% of the total plan expenditures) based on lifetime contributions. Both contributions and benefit accruals stop once earnings get to a level approximately equal to the Average Wage. Thus, the benefits are not enough for a comfortable retirement for most Canadians. OAS plus C/QPP benefits replace about 40% of earnings for an average worker. C/QPP benefits are taxable income, but there is no claw back.

10

Figure 1

Canadian Security System in 2006-Total Benefits

0

3

6

9

12

15

18

21

1 6 11 16 21 26 31 36 41 46 51 56 61 66 71 76 81 86 91 96

Yearly Preretirement Income (in thousands of Canadian Dollars)

Yea

rly T

ota

l R

etir

emen

t B

enef

its

(in

th

ou

san

ds

of

Can

ad

ian

Doll

ars

)

Net OAS CPP GIS

Figure 1 indicates quite clearly that if one were to depend totally on the government-sponsored systems for one’s total retirement income, then a wide range of Canadians would live on virtually the same income post-retirement. One can see here (and later in Table 1) that the Canadian government-sponsored retirement income system is heavily targeted to the poor and much less concerned about individual equity. Figure 1 and Table 1 assume that if the government does not provide the worker with a 70% replacement ratio, then the worker will save privately to bring retirement income to that level. Figure 1 shows how rapidly the GIS benefits are “clawed back” with its 50% reduction factor. For many Canadians the reality is even more stark. This is because some provinces (notably Ontario which has the largest provincial population) also pay welfare benefits to the elderly. Thus, for an Ontario citizen, one loses one’s total welfare benefits (federal plus provincial GAINS) $1 for $1. Hence, it makes very little sense for most poor Canadians to save for retirement. This is a negative side effect that must be noted and understood in the design of any social security system. The maximum benefit of the C/QPP will only replace 25% of earnings up to the Average Wage. Thus, there remain strong incentives for all but poor Canadians to continue to save for retirement

11

within the private sector and for workers to strive for workplace pensions. In this regard, Canada offers significant tax incentives to save for retirement. Within limits, and whether done as an individual or through a workplace pension, contributions (both employer and worker) are tax deductible, investment income accrues tax free, but benefits are income taxable (in full) when taken. There are also limits as to when you must start to take your retirement income and both a minimum and maximum amount that must be taken in any year. Annuitization is not mandated.

12

Figures 2 and 3 confirm just how focused the Canadian system is on the alleviation of poverty. All of the GIS benefit and much of the OAS benefit is targeted solely to the poor. This will be discussed more fully later in the paper. As can be seen in Figure 4, the U.S. uses a two-pronged attack. First, there is a very small welfare scheme referred to as Supplemental Social Insurance (SSI—the black area) which is financed by general tax revenues. One can see that this is a minimalist program with a very sharp “claw back” of benefits. You lose one dollar of your SSI for every dollar of private income you have after some small exempt amounts (see http://www.ssa.gov/policy/docs/statcomps/oasdi_monthly/ for more details). Again, it should be noted that this probably discourages poor Americans from saving privately for retirement (but given the minimal size of the SSI, only very poor Americans)

13

Because SSI is so minimalist, OASDI has acquired a multiple personality. First, it must alleviate poverty for many of the poor elderly because SSI, by itself, will not do that. Second, it must create a replacement ratio that satisfies the need for some predictable standard of living post-retirement. Plus, it must create solidarity amongst all participants (which requires benefits for wealthier participants). The OASDI system attempts to be all things to all people through a three-part benefit formula (see www.ssa.gov/OACT/COLA/Benefits.html). OASDI also has stronger Survivor benefits than many other Social Security systems. This is probably a result of the timing of the design of the original OASDI system in the late 1930’s. At that time poverty was a serious problem for many of the U.S. elderly, but was especially difficult for widows and females living alone. Thus, the original architects made sure that the spouses (normally wives) of OASDI worker participants got significant continuing benefits when the worker passed away. Figures 5 and 6 (which follow) can be used to compare the level of

14

targeting of benefits in the U.S. versus Canada (as in Figures 2 and 3).

15

Table 1 Ratio of "Income Replacement Ratios of the U.S. System" to "Income Replacement Ratios of the

Canadian System" in 2006

Pre-Retirement Income Ratio $1,000 0.57416 $5,000 0.54786

$10,000 0.54417 $15,000 0.62875 $20,000 0.70652 $25,000 0.84952 $30,000 1.06107 $35,000 1.07981 $40,000 1.09560 $45,000 1.15818 $50,000 1.22952 $60,000 1.32112 $70,000 1.41272 $80,000 1.50432 $90,000 1.61078 $100,000 1.78556

What one can sense from the previous Figures is now confirmed in Table 1. That is, the Canadian system is highly focused on poverty alleviation and less concerned about individual equity. The U.S. system is much less focused on alleviation of poverty and much stronger in terms of the principle of individual equity (benefits are a function of contributions). This will be discussed more in the next section. The focus on “equity” versus “adequacy” has other effects. One might expect to experience more “solidarity” toward OASDI in the U.S. than to the C/QPP in Canada since the wealthy get more for their contributions in the U.S. than do similar workers in Canada. On the other hand (as will be seen later) the Canadian system will probably do more to alleviate poverty than the system in the U.S. Referring to Figure 1 (Canada) and Figure 4 (U.S.) it is clear that the total area under the curves represents the “cost” of Social Security (at least the Retirement Income portion) in the two countries. In general, this cost can be paid by general tax revenues, earmarked contributions and investment income. Despite the fact that some portions are financed by general taxation, any designer of Social Security must be cognizant of the fact that workers will only pay so much for

16

income security for the elderly. This is true even if part of the cost is partially “hidden” in general taxation income. How could one lower the overall cost of these systems? First, one could make the welfare benefits (GIS/SSI) smaller by having a faster (steeper) claw back. However, given that we are already subjecting our poorest elderly to effective tax rates of 50 percent and above, this seems an unfair target. Further, this does not seem prudent since it discourages poorer workers from saving for retirement. As an alternative, one could lower the benefits paid to the wealthiest citizens. Surely they can fend for themselves if the government cuts back their publicly-funded protection. That may be possible, but it would also erode the solidarity inherent in the overall system. Would the wealthy continue to support the program if they reaped much smaller benefits? These are difficult questions and compromises will be required. One also wishes to assure the public that the cost/benefit structure is sustainable. This is normally done through actuarial evaluations of the system that project many years into the future (e.g, 75). In this regard, both OASDI and the QPP face some long-term sustainability issues. VI Maximum Benefits and the Impact on the Labor Force In Canada, within the C/QPP, to achieve maximum benefits, you must have full contributions for a number of years established by the formula: 0.85(65-18). That is, you must have made full contributions for 85% of the years between age 18 and age 65. This works out to 40 years of required contributions. There are some exceptions to this general rule. First, you can drop out years (within the bracket) when you are home, caring for children and years when you are collecting C/QPP Disability Income. The child-rearing drop-out requires that the child be under age 7 (verifiable with government birth records) and the carer’s income does not exceed a set level (verifiable through income tax records). This creates a small incentive for parents to leave the work force for a period of time when they have small children. Apparently there exists no literature that attempts to determine the overall impact of this feature on labor force participation rates. However, as stated before, an explicit drop out formula should be preferred to an overall short qualification period for full or significant benefits. Another issue is the adjustment used if you retire early. Canadians can retire as early as age 60 and take their C/QPP benefits, but will incur an actuarial reduction of benefits of 0.5% a month. Thus, a worker who retires at age 60 will only receive 70% of the income they would have received at age 65. On the other hand, if you delay retirement until age 70 (the upper bound) a worker will receive a 130% benefit. (A CPP study states that the 0.5% adjustment is not a full actuarial adjustment and should be larger. The error gets larger as the age of retirement

17

advances. This creates a small incentive for earlier, rather than later retirement (Office of the Chief Actuary, 2003)). Another important issue is raised here. Are we better to have an “early retirement factor” equal to ½% a month (6% a year) that is easy to digest and remember or are we better to have “correct” actuarial adjustments (perhaps to several decimal places) that no one can comprehend or remember? The old adage “Keep It Simple, Stupid” may apply to these public plans. But that is more generally true. To enhance transparency and comprehension, the overall benefit design of social security must be kept as simple as possible. If workers cannot comprehend the benefit structure, then attempts to affect changes in attitudes towards saving for retirement and age-at-retirement will inevitably fail. Regardless of when one retires, there is very little benefit in the C/QPP retirement income one receives in working beyond 40 years (although it is possible to substitute years of higher income for previous years of lower income). This is especially true because, once you retire, you never make any more CPP contributions even if you go back to work (you do in the QPP). Thus, the C/QPP creates an incentive to leave the labor force after 40 years (and age 60). Internationally, this is actually one of the largest qualification periods used for this criterion. However, this is another important variable to consider in designing any social security system. That is, have you created incentives that discourage labor force participation? Some countries use requirements less than 40 years to implicitly allow for time out of the labor force to care for children, the disabled or the elderly. As stated before, it is preferable to establish an explicit allowance (such as in Canada) especially when it can be verified without a large bureaucracy (again as in Canada) than to have a shorter contribution or participation requirement for all. In the U.S., maximum benefits under OASDI can be achieved after 35 years of contributions (later years of higher earnings can be used to substitute for years of lower earnings if participation exceeds 35 years, however). Many papers have been written that indicate that this creates incentives for workers to leave the labor force after 35 years (see Shah et al, 2006). VII Measures of Success How can one measure the success of a Retirement Income Security system? In two recent papers, Brown and Prus (2004, 2006) attempted to analyze the balance between adequacy and equity in the Retirement Income Security programs of several developed countries. These are competing goals. The more strongly the system ties benefits to contributions (resulting in individual equity) the less redistribution of wealth is available making the achievement of “adequacy” goals more difficult. Other papers on this topic in the actuarial literature include: Brown and Ip (2000), Knox and Cornish (1997). The Brown/Prus paper adds to the “adequacy and equity” criteria, the goal of attaining high

18

levels of security while still allowing for a strong private sector presence in the attainment of retirement income security. While this may not be universally accepted, most actuaries would find this a laudable goal. In the first of their two papers, Brown and Prus (2004) juxtapose the level of post-retirement income disparity against the amount of retirement income (as a percentage of total income) provided from government sources. Post-retirement income disparity is measured by the Gini index, a statistic whose value lies between 0.0 (completely equal income distribution) and 1.0 (one individual holds all income). The authors find a negative correlation between income disparity and the size of the government transfer program. In a ten-country study, the authors find that Sweden has the lowest Gini coefficient (0.194) and the government provides 69.6 percent of post-retirement income. At the other extreme analyzed, Israel has a Gini index of 0.374 with government transfers representing 30.5 percent of retirement income. How do Canada and the U.S. fare? Canada has a Gini index of 0.256 and government transfers equal to 46.3 percent of post-retirement income (standing fifth on both statistics in the ten-country sample). The U.S. has a Gini index of 0.364 with the government providing 39.1 percent of retirement income (a ninth place standing on both statistics). Other research has arrived at similar conclusions (e.g., Regie des rentes du Quebec, 2004; OECD, 2000; 2001). The 2001 OECD study show how targeted the Canadian social security system is to the alleviation of poverty among the poor versus stronger individual equity principles in the U.S. system. Table 2: Disposable income of the population aged 65 and over by Income Decile compared with the Population aged 18 to 64 in the same Income Decile, mid 1990s Decile Canada US

1 148 80 2 107 78 3 94 77 4 87 78 5 85 78 6 86 81 7 86 83 8 86 94 9 87 83

10 96 94 OECD 2001, p21

19

VIII Conclusion The paper suggests that policy-makers (and their economic and actuarial advisors) when designing (or re-designing) a social security system, need to keep all of the following principles in mind: • Benefits will be paid for by a combination of taxes, contributions and investment income. Regardless of the financing source, the total cost of the system must be affordable and sustainable. In this paper, the “cost” of the benefits can be viewed as the area under the appropriate curves. • The system and its inner workings should be understandable and comprehensible to a large proportion of the population. This also supports the goal of transparency. • It is preferable to have explicit drop-out provisions for contingencies such as caring for children or disability than to offer a significant benefit with only a short work-force attachment. • The benefit/contribution structure should not encourage workers to evade participation by not admitting to earnings or by entering the cash economy. • The removal of welfare benefits as participants enter higher wealth zones should not be so rapid as to create large effective marginal tax rates that will then create incentives so that workers will not save for retirement. • It may be necessary to provide significant benefits to wealthy participants to guarantee the solidarity of these workers in the support of the system as a whole. • The benefit/contribution structure should not encourage workers to leave the work force early. For example, a 40-year work history is required in Canada before full benefits can be paid. These are all very important principles. They are also often in conflict among themselves. It will be extremely difficult to achieve all of the goals outlined above (e.g., total cost versus steep claw backs). One must also be cognizant and sensitive to the local culture and history in designing social security pension systems.

20

Bibliography Barr, N. 2000. Reforming pensions: Myths, truths, and policy choices. Working Paper

WP/00/139, International Monetary Fund. Washington, DC. _________ 2002. “Notional defined contribution pensions: What they can do, and what they

can’t.” Scandinavian Insurance Quarterly, Issue Nr 3. Brown, R. L. 1997. “In defense of pay-as-you-go (PAYGO) financing of social security.”

North American Actuarial Journal. Vol. 1, No. 4, October, 1997, pp 1-20. Brown, R. L. and J. Ip. 2000. “Social security: Adequacy, equity, and progressiveness: A

review of criteria based on experience in Canada and the United States.” North American Actuarial Journal 2: 1-33.

Brown, R. L. and S. G. Prus. 2004. “Social transfers and income inequality in old age: A

multinational perspective.” North American Actuarial Journal 8: 30-36. ___________________________ 2006. “Income inequality over the later-life course: A

comparative analysis of seven OECD countries.” Annals of Actuarial Science, Vol. 1, No. 2, pp. 307-317.

Cichon, M. 1999. “Notional defined-contribution schemes: old wine in new bottles?”

International Social Security Review, Vol. 52, Pt. 4, pp 87-105. ISSA, Geneva. _________ 2005. “Balanced notional defined contribution schemes: A new “geist” in old

bottles?” Scandinavian Insurance Quarterly. Issue Nr 2. Conesa, J. C. and C. Garriga. 2004. Optimal design of social security reforms. Hagberg, J. and E. Wohlner 2002. “The market for social insecurity: A shady pension reform

entices economic elites.” Scandinavian Insurance Quarterly. Issue Nr 4. International Actuarial Association. 2006. “Comments on the ‘old-age income support in the

21st Century’ report released by the World Bank in February 2005.” www.actuaries.org/LIBRARY/Submissions/SOCSEC_World_Bank_final.pdf

Knox, D. and R. Cornish. 1997. The Development of some criteria for equity in national

retirement income systems. Working Paper Series. University of Melbourne, Australia. Office of the Chief Actuary. 2003. Canada pension plan actuarial adjustment factors study.

Actuarial study No. 2. March. Office of the superintendent of financial institutions.

21

Organisation for Economic Co-operation and Development (OECD). 2001. Ageing and income: financial resources and retirement in 9 OECD countries. Paris.

_________ 2000. Reforms for an ageing society. Paris. Regie des rentes du Quebec. 2004. Evaluation of the Quebec system of financial security at

retirement in relation to that of other industrialized countries. Province of Quebec. Scherman, K. G. 2003. “The Swedish pension reform: a good model for other countries?”

Scandinavian Insurance Quarterly. Issue Nr 4. _______________ (2005): A review of the Swedish pension reform: Strengths and weaknesses.

Contribution to the 10th International Pension Research Institute, November 22, Tokyo, Japan.

Shah, G., J. Shoven, S. N. Slavov. 2006. Removing disincentives for long careers in social

security. Presented at the Society of Actuaries “Re-envisioning retirement in the 21st century”, Washington, D.C. May. http://www.soa.org/ccm/content/research-publications/library-publications/monographs/retirement-systems-monographs/

World Bank. 2005. Old-age income support in the twenty-first century: International perspective

on pension systems and reform. February. http://info.worldbank.org/etools/docs/library/139481/HolzmannHinz_OldAgeSupport.pdf